Meet S. Kevin Zhou, MICCAI Fellow 2022

S. Kevin Zhou was named a MICCAI Fellow in 2022 in recognition for his outstanding contributions to the theory and application of medical image computing and their translation into clinical practice. His research is dedicated to advancing medical image analysis and reconstruction, and its applications in real practices.

Kevin has been an active leader in the MICCAI community, having been elected to the MICCAI Society board of directors in 2019 and serving as Treasurer from 2021-2024. He also made a significant impact as Program Co-Chair of the MICCAI 2020 conference and was actively involved in organizing multiple MICCAI conferences and workshops.

Beyond MICCAI, Kevin plays a key role in the broader medical imaging field. He serves on the advisory board of the Medical Open Network for AI (MONAI) project (2020-2024) and holds editorial positions with leading journals, including Medical Image Analysis, npj Digital Medicine,IEEE Transactions on Pattern Analysis and Machine Intelligence, and IEEE Transactions on Medical Imaging (2016-2023). His contributions have earned him fellowships with AIMBE, IAMBE, IEEE, MICCAI, and NAI.

Kevin is currently a Distinguished Professor and Executive Dean of the School of Biomedical Engineering at the University of Science and Technology of China (USTC). He is also an adjunct professor with the Institute of Computing Technology, Chinese Academy of Sciences and Chinese University of Hong Kong. He directs the Jiangsu Provincial Key Laboratory of Multimodal Digital Twin Technology, and the Center for Medical Imaging, Robotics, Analytic Computing & Learning (MIRACLE) at USTC Suzhou Institute for Advanced Research.

Prior to his academic appointments, he was a Principal Expert and Senior R&D Director with Siemens Healthineers Research (formerly Siemens Corporate Research). He has published more than 350 book chapters, peer-reviewed journal articles and conference papers and holds more than 150 patents. He received his PhD from the University of Maryland, College Park; his master’s degree from the National University of Singapore; and his bachelor’s degree from the University of Science and Technology of China.

We celebrate Kevin’s remarkable impact on medical image computing, computer vision, and his enduring contributions to the MICCAI community. Get to know more about his journey, insights and experiences in our exclusive Q&A interview.

Q. Can you tell us about your journey into the field of medical image computing and computer vision? What first drew you to this area of research?

A. I entered the field of computer vision when I was a graduate student, working on texture perception and face recognition. I later joined Siemens Corporate Research and focused on medical image computing, specifically researching innovative machine learning approaches for effective, efficient, and automated medical image analysis. During that time, it was extremely challenging to develop a product-quality algorithm, and my strong desire was to do so.

Q. You've published extensively (in more than 350 peer-reviewed journals, conference and workshop papers) and have been cited widely (more than 23,000 citations). Of all your publications or research papers, is there one that stands out as particularly meaningful or groundbreaking to you? 

A. Shape regression machine (SRM) and its variants, which enabled segmentation of heart, left and right lungs, liver, left and right kidneys, and spleen with an accuracy comparable to an expert’s processing a 3D volume of hundreds of slices, all in a lightning speed within just 1-2 seconds in a 3DCT on a PC using CPU computation. Note that this was achieved before the deep learning era. Such a strong performance arises from a novel algorithm design, which treats a segmentation problem as a regression problem for the first time in the field, and an extensive engineering optimization (courtesy to my Siemens colleagues). Later, this algorithm has become a part of a product that helps millions of patients.

Q. More than 150 patents have been granted and thousands of hospitals around the world use products based on your algorithm. How do you decide whether a research idea is worth pursuing toward a patent or commercialization?

A. I believe that this should be based on clinical value, which is often figured out via thoughtful discussions with physicians. Take rib unfolding for example. It first detects the rib center-line voxels using a learned binary classifier and then adapts a 3D rib cage model using belief propagation with robustness, accuracy, and efficiency. The patient-specific rib cage model is then unfolded such that 3D ribs are flattened onto a 2D plane, providing a holistic overview of 24 labeled spine vertebrae and ribs. Rather than having a radiologist examine each pair of curved and twisted ribs in the slices along the head-to-toe axis, such an unfolded picture enables radiologists to read a thoracic CT scan with clarity and efficiency. The automated labeling also allows direct identification of ribs of interest for reporting. This feature has been used primarily in oncology for detecting bone metastases and in acute care for detecting bone fractures, where it increases diagnostic sensitivity by 10%. Furthermore, it reduces the reading time for bones by 50% as compared to conventional reading.

Q. What’s one piece of advice you wish you’d received early in your career?

A. The importance of teamwork. “There is no limit to what a man can do or where he can go if he doesn't mind who gets the credit.”

Q. Looking ahead, what’s next for you? Are there any projects or initiatives coming up that you’re particularly excited about?

A. I am particularly excited about the idea of building a general-purpose AI solution for a radiologist. The current AI solution is of narrow focus, supporting a curated list of diseases and modalities. My goal is to address all diseases and all modalities. Another initiative I am taking is to build a so-called Medical Time Machine (MedTM), which can predict the longitudinal evolution of medical images based on the current observations and hence better facilitate the decision-making of future health conditions.

Q. How do you see MICCAI’s role evolving in the coming decade, especially with the rapid development of AI in healthcare?

A. I envision that MICCAI will and should play a leading role in the coming decade, especially if MICCAI can collaborate closely with healthcare stakeholders.